TrigInDetAnalysis package
General description
The TrigInDetAnalysis package is devoted to the study and analysis of the performance of LVL2 tracking algorithms.
Two ROOT based macros are currently available:
Efficiency.cpp: evaluates efficiency for the tracking algorithms and the amount of fake reconstructed tracks as a
function of η and pT; it also provides overall efficiency values and an average track multiplicities per event as a
function of the number of linked space points requested for the good tracks.
Resolution.cpp: evaluates track parameters resolution (η, φ, z0, pT, d0)
using the reconstructed tracks linked to a kine by a minimum number of space points.
Variables definitions
The meaning of the quantities we refer to when describing the plots produced by the analysis macros is here summarized as follows:
Reconstructible kine track: a kine track passing all the geometrical selection cuts (being contained in one of the
RoIs, crossing the pixel B-layer, being a primary track) and an optional pT cut;
Good track: a reconstructed track linked to a good kine by at least n space points, where n is a user
selectable integer number ranging from 1 to the maximum number of space points used to build a track; the geometrical and
pT cuts used for the kine apply also to the reconstructed tracks;
Best track: for each reconstructible kine, more than one good track can be available; we define the best track as
the one with the most precise η-φ matching w.r.t. the linked kine;
Fake track: a reconstructed track (passing geometrical and pT cuts) which is not a good track;
Efficiency: ratio between best tracks and reconstructible kines;
Fake fraction: ratio between fake tracks and all the recostructed tracks passing the applied cuts;
Track multiplicity: the average number of reconstructed tracks per event passing the applied cuts.
Plot description
Each analysis macro produces a set of histograms; their are here briefly described.
Efficiency.cpp: by default good tracks must have at least two kine hits;
top: the efficiency is plotted as a function of pT and η; for the η plot the optional
pT cut is applied;
center: the fake fraction is plotted as a function of pT and η;
bottom: the left plot shows the overall efficiency (integrated over η and pT, above the optional cut)
as a function of the minimum number of kine hits required for a good track; the right plot shows the average track multiplicity
as a function of the minimum number of kine hits, so that the first bin corresponds to the average multiplicity for all the
recostructed tracks.
Resolution.cpp: by default resolution plots are filled using the best tracks with at least two kine hits; optionally
is possible to apply a pt cut, to vary the minimum number of kine hits for good tracks and to fill the plots with all the good
tracks and not only with the best ones.
Moving from left to right and from top to bottom the parameters displayed are: η, φ, z0, 1/pT,
d0.
Instructions for running the analysis macros
Both the analysis tools take three strings as input arguments, like in
./efficiency DatasetLabel AlgoritmList [Options]
where DatasetLabel defines the set of ROOT files to be analyzed, AlgorithmList specifies the list of algorithms to be
compared in the plots and the optional parameter Options sets up the available running options.
Specifying the dataset
The dataset to be analyzed is selected using a case-insensitive label which must be defined in the datasets.dat file. For
each dataset it must contain an entry like
BEGIN_LABEL DatasetLabel
PARAMETER ParameterValue
PARAMETER ParameterValue
...
END_LABEL
where DatasetLabel is the label used to tag a particular set of ROOT files, while the parameters, specifying the place where
data files physically reside, are
PATH: the directory where the ROOT files are stored; if the directory lays in a remote file system (like CASTOR) the
string must begin with rfio:; example:
rfio:/castor/cern.ch/user/b/baines/trigprod/rel702++2003-12-12/lumi02/002026/his/;
FILE: the names of files sequentially produced with the standard job submission tools are structured like
description_number.root; the FILE parameter specifies the description part of the filename; example:
ntuple_e25_lumi02_002026_;
FIRST: selects the first file to be processed within the sequential list selected by the FILE parameter;
LAST: selects the last file to be processed within the sequential list selected by the FILE parameter;
TREE: specifies the name and path of the ROOT tree where CBNT info is stored inside the file; the default value is set to
CBNT/t3333.
Selecting the algorithms
The list of algorithms to be compared is passed to the macro as a single string; the names of the various algorithms can be
separated by any character (or even no character at all) and the recognition is case-insensitive; the only valid algorithm names
are:
IdScanOff, SiTrack, IdScanOn, TRTXk.
Setting the available options
The running options are passed to the macros as a string composed by KEY = value; blocks where KEY identifies a
particular option, the equal sign links the KEY field to its selected value and the semicolon is the separator between different
options.
Here follows the list of valid option keys for the two analysis tools.
Efficiency: the options and their meaning for this macro are:
PtCut: a float number which corresponds to the optional pT cut applied; default value: 0.0;
MinKine: the minimum number of kine hits needed to tag a reconstructed track as a good track; default value:
2;
Width: display window width; automatic width is mantained selecting -1; default value: -1;
Height: display window height; automatic height is mantained selecting -1; default value: -1;
Print: boolean flag enabling or disabling eps and gif files production; if this option is enabled, the
plot filenames are built as follows: eff-DatasetLabel-PtCut.eps [.gif]; default value: false.
Resolution: the options and their meaning for this macro are:
PtCut: a float number which corresponds to the optional pT cut applied; default value: 0.0;
MinKine: the minimum number of kine hits needed to tag a reconstructed track as a good track; default value:
2;
Width: display window width; automatic width is mantained selecting -1; default value: -1;
Height: display window height; automatic height is mantained selecting -1; default value: -1;
UseBest: boolean flag choosing wether resolutions are evaluated using all the good tracks or only the
best ones; default value: true;
Print: boolean flag enabling or disabling eps and gif files production; if this option is enabled, the
plot filenames are built as follows: res-DatasetLabel-PtCut.eps [.gif]; default value: false.
Results
Some plots produced with the TrigInDetAnalysis package are available on the
LVL2 tracking algorithms performance page.
Contact person: Carlo Schiavi